Department of Physics and Astronomy
DIGITAL IMAGE PROCESSING
Course 3624
Topic 5 - Imaging Mapping - II
Professor Bob Warwick
5.1
The Perception of Colour
• Although the human eye has only a limited ability to discern
shades of grey (~ 2 5), it does much better with colour
gradations (> 2 8)
• Inclusion of colour invariably adds to the visual content of an
image (eg black and white versus colour TV pictures)
• The human perception of colour is a complex subject. It
depends on both physiological and psychological factors. Also
the eye/brain system is highly adaptive
The Nature of White Light
Colour Mixing Experiments
Combining colour pigments
Magenta, cyan, yellow, and black are
the colours often used in printing
The Human Eye
The Retina
Fovea
Macula
5
Optic nerve
Spectral Sensitivity Curves of Rods and Cones
The range of sensitivity is roughly 400 - 650 nm.
There are three types of cone (blue, green, red).
The rods (dashed) primarily give sensitivity in low illumination
Tri-Stimulus Colour Theory
Any colour can be reproduced by mixing an appropriate
set of three “primary colours” – Thomas Young 1802
A More Complex Model of
Human Colour Perception
Colour Matching Experiments
Choose three “primary” light sources with spectra
P1(λ), P2(λ) & P3(λ)
C(λ) =
Colour matching involves adjusting the input light levels
β1 β2 β3 to match the tristimulus values of the test colour
Colour Matching with Monochromatic
Primaries
435.8 nm
546.1 nm
700nm
The CIE 1931 (All Positive) Colour
Matching Functions
The corresponding tristimulus
values for an arbitrary spectral
colour I are:
The CIE XYZ system was designed so that the Y parameter was a
measure of brightness ie luminance. Then
x=X/(X+Y+Z) & y=Y/(X+Y+Z) represent "chromaticity" parameters,
(since z = 1 - x – y gives no further colour information)
Thus three parameters define an input in the CIE xyY colour space.
CIE 1931 Chromaticity Diagram
The outer curve boundary represents
the track of monochromatic colours in
the x,y plane. (Although in practice the
colours shown are restricted by the
performance of the display devices.)
The gamut of colours that can be
generated using monochromatic
primaries at 435.8, 546.1 and 700 nm,
lie within the triangle. Similarly the xy
coordinates of the three phosphors in a
display will define the "realizable"
colours of the device
Alternative Colour Specifications
Colour shade (red, green,blue etc..)
Depends on peak λ
Measures the purity of the colour
Degree of dilution by white light
Luminance Hue Saturation (LHS) Colour Space
Complementary Colours
CYAN - RED
GREEN - MAGENTA
BLUE - YELLOW
Colours opposite each other on the colour disk are called “complementary”.
Author: Richard Alan Peters II
5.2 Pseudo-Colour Techniques
Colour displays utilize three different phosphors (RGB) to generate
a set of "realizable" colours. These colours occupy the "colour
cube" of the device, which can be visualized as a 3-d space.
B
R
G
The Colour Cube has R,G & B axes normalised so that at full intensity
R=G=B=1.
Then R=G=B=0 is black
R=G=B=1 is full intensity white
R=G=B= 0  1 represents a grayscale (along the leading diagonal)
Pseudo-Colour Processing
Pseudo-colour processing involves the use of colour to enhanced the
display of a single band (monochrome) image. The input grayscale is
subdivided into intervals, against which a set of colours are assigned.
If the number of intervals
is small  Density Slicing
Colour
Assignment
Yellow
Red
Blue
Black
Grey Level ( f )
The Colour Table
More sophisticated applications of pseudo-colour involve the use of
a colour table. The Colour Table represents a specific (and more
systematic) mapping from grey level to RGB colour.
For example:
R,G,B
1
0
0
63
127
191
255
f
Example: Colour Table with 64 levels
A single band image in pseudo-colour.
XMM Image of SNR G21.5-0.9
5.3 True Colour and False Colour
Techniques
In many applications (e.g., remote sensing, astronomy etc..) more
than one image is recorded of the same scene using different spectral
filters. Up to three such images can be displayed simultaneously
with a "three channel" display system.
In this case the possibilities are:
TRUE COLOUR PROCESSING - where the images are recorded
through R,G & B filters and are (respectively). displayed on the R, G
and B channels of the display system. In many applications (e.g.
photography, TV etc..) special care is needed to achieve an
acceptable level of colour fidelity.
FALSE COLOUR PROCESSING - where images recorded in three
different wavebands are displayed arbitrarily on the R, G and B
channels of the display system.
Implementation of Colour Enhancement
via the device Colour (Look-up) Table
Single Channel System – Pseudo-Colour
Image
Store
Colour
Look-up
Table
R
D/A
G
D/A
B
D/A
Video
Out
Three Channel System – True or False Colour
Image
Store 1
Colour
Look-up
Table
R
D/A
Image
Store 2
Colour
Look-up
Table
G
D/A
Image
Store 3
Colour
Look-up
Table
B
D/A
Video
Out
Video
Out
Video
Out
True Colour
DISPLAY
I/P CHANNELS
False Colour
DISPLAY
I/P CHANNELS
10 micron image
2.2 micron image
1.2 micron image
Pseudo Colour
DISPLAY
I/P CHANNELS
Pseudo Colour Examples
True, False or Pseudo
Colour ?
Studying Temporal Changes in Remote Sensing
Example: Panama Canal